As the information age arrives, intelligent buildings develop rapidly. People are more critical to quality of elevator service, thus, only a single elevator cannot meet the traffic requirement of a building. In this way, multiple elevators are to be fixed reasonably; Elevator Group Control System (EGCS for short) ensues. EGCS, together with a central controller, collects operational information of every elevator then deliver control signals to each one to improve the quality of service and reduce energy consumption. Group control of elevator system deals with strategic problems of multiple-objectives and random system that is complicated, non-linear and uncertain. Therefore, it is essential to combine knowledge and information of fuzzy control, neural network and real time embedded system to cope with elevator group control system.This paper starts from dynamic features of elevator, and then analyzes variables and traffic features of elevator group. In accordance with different control expectations, we set up corresponding objective functions and restriction formulas. Compared with traditional and current used arithmetic, we put forward FCMAC (Fuzzy Cerebellar Model Articulation Controller) and provide math testification and calculation method. Group control method of fuzzy neural network takes average waiting time, energy consumption and average service time as its value functions. This method control energy effectively and insure the satisfaction rate of passengers. FCMAC can respond to passenger flow and other real time status, balance the capability of elevator, waiting time and support online learning.Elevator has a high level of real time performance and requires more expansion interfaces. We make TMS320F2812 from TI as the core of hardware design to realize the connection of RS-232, RS-485, RS-422, USB and stability of electrical source. Moreover, the design offers additional interfaces for future use on communication aspect.
|